mesoscale modeling of dislocation climb and primary creep …€¦ · mesoscale modeling of...
TRANSCRIPT
Mesoscale modeling of dislocation climb and primary
creep process in single crystal Ni base superalloys
S.M. Hafez Haghighat1, Z. Li1, S. Zaefferer1, R.C. Reed2, D. Raabe1
1Max-Planck-Institut für Eisenforschung, Max-Planck-Str. 1, 40237 Düsseldorf, Germany
2Department of Engineering Science, University of Oxford, Parks Road, Oxford, OX1 3PJ, UK
International Workshop on Dislocation Dynamics Simulations
Dec. 10-12, 2014, Saclay-France
Ni base superalloys
• Single crystal Ni base superalloys:
• Used in the blades of advanced gas turbines
• Operating at elevated temperatures (~1000°C)
• Microstructure:
• γ´ particles of Ni3Al (L12) oriented along <100>
• Separated by thin γ channels of fcc-Ni alloy
• Creep mechanisms:
• Filling of γ channels by dislocations
• Formation of dislocation networks on γ/γ´ interfaces
• Cutting of γ´ particles by dislocations
• Rafting process
• At early stages of high temperature creep the main
controlling mechanisms are the dislocation processes,
glide and climb.
• Discrete dislocation dynamics (DDD) can give further
insight into the microstructure of Ni base superalloys
during creep.
1
CMSX-4
Dislocation climb rate
• Dislocation mobility law for DDD:
2
𝑽 =𝑭
𝑩⟹
𝑽𝑔 =𝑭𝑔
𝐵𝑔
𝑽𝒄 =𝑭𝒄
𝐵𝑐⟹ 𝑽𝒄 =
𝒇𝑐𝑃𝐾 + 𝒇𝑜𝑠
𝐵𝑐
𝒇𝑐𝑃𝐾 = 𝝈. 𝒃 × 𝒕 ∙ 𝒏 𝒏
𝒇𝑜𝑠 = 𝜇𝒃
Ω
𝒏
𝒏=
𝐾𝐵𝑇 𝒃
Ω
𝒏
𝒏ln(
𝑐
𝑐0)
At equilibrium: 𝑭𝑐 = 0 ⟹ 𝑐𝑒𝑞 = 𝑐0exp −𝒇𝑐
𝑃𝐾Ω
𝑏𝐾𝐵𝑇
Dislocation climb rate
• Vacancy current to/from dislocation core:
3
𝐼𝑣 = −2𝜋𝑟𝑐𝐾(𝑐𝑒𝑞 − 𝑐)
𝑣𝑐 = 𝐼𝑣Ω/𝑏 ⟹ 𝑣𝑐 = −𝜂
𝐷
𝑏𝑐0exp −
𝑓𝒄Ω
𝑏𝐾𝐵𝑇 − 𝑐
𝜂 =2𝜋Ω
ln𝑙𝑏
=2𝜋Ω
ln (1
𝑏 𝜌𝑑) 𝑐0 = exp −
𝐸𝑓𝑣
𝐾𝐵𝑇 𝐷 = 𝐷0
𝑒𝑓𝑓exp −
𝑄𝒆𝒇𝒇
𝑅𝑇
𝑣𝑐 = 𝑓(𝑇, 𝑓𝒄, 𝜌𝑑, 𝑐, 𝑐0 𝐸𝑓𝑣, 𝑇 , 𝐷(𝑇, 𝐶𝑜𝑚. ))
• Mordehai et al. Phil. Mag. 2008
• Shyam et al. PRL 2013
• Zhu et al. Acta Mat. 2012
• Efv = 1.4 eV
• Com.; CMSX-4
A recent model
• Danas and Deshpande, MSMSE 2013:
But not dependent on c
4
𝑣𝑐 = 𝑓(𝑇, 𝑓𝒄, 𝜌𝑑, 𝑐0 𝐸𝑓𝑣, 𝑇 , 𝐷(𝑇, 𝐶𝑜𝑚. ))
⟹ 𝑣𝑐 =2𝜋𝐷𝑐0Ω
𝑏2𝐾𝐵𝑇ln (1
𝑏 𝜌𝑑)
𝑓𝑐
𝐵 =
𝑏2𝐾𝐵𝑇ln (1
𝑏 𝜌𝑑)
2𝜋𝐷𝑐0Ω
𝑐 = 𝑐0
𝑓𝒄Ω ≪ 𝑏𝐾𝐵𝑇
Comparing the two equations
• Eq.(1):
• Eq.(2):
5
𝑣𝑐 = −𝜂𝐷
𝑏𝑐0exp −
𝑓𝒄Ω
𝑏𝐾𝐵𝑇 − 𝑐
𝑣𝑐 =2𝜋𝐷𝑐0Ω
𝑏2𝐾𝐵𝑇ln (1
𝑏 𝜌𝑑)
𝑓𝑐
Parametric variation of climb velocity
Climb mobility dependence:
1. Temperature
2. Applied stress
3. Vacancy concentration
4. Dislocation density
6
Constant Bc/Bg ratio for glide-climb
mobility during a simulation.
Discrete dislocation dynamics method
• Defining a simulation box with periodic boundary conditions (PBC).
• Dislocations are randomly located in the simulation box and decomposed into
straight segments having character from pure edge to pure screw.
• Stress on the dislocation segments is calculated through the Peach-Koehler
forces due to applied stress and the stress field of other dislocations.
• The position of segments is defined by calculating the speed of the segments
(F=BV) and local events occurring during the displacement, such as direct
annihilation, junction formation, cross-slip, climb and interaction with particles.
7
dat
tpk ubF
)(
8
Hybrid mobility law
• To investigate the specific effect of climb:
o ParaDiS code
o Glide-climb dislocation mobility law
• Proposed glide-climb mobility of a dislocation in Ni
base superalloys:
• Hybrid mobility law:
Z. Zhu et al., Acta Materialia
60 (2012) 4888–4900
b, n and t are unity vectors of Burgers
vector, glide plane normal and
dislocation line direction.
)()( nnBmmBB cggc
)( nnBB coc tnm
2/12222 ])([ tbBtbBB sgegg
tbBB ecc
S.M. Hafez Haghighat, G.
Eggeler, D. Raabe, Acta
Materialia 61(2013) 3709.
9
Dislocation - ’ particle interaction
In the above, there are three stages:
A. Easy glide of dislocation before approaching the ’ particle
B. Dislocation slip on the interface by the glide/climb mobilities and deposition of the lateral
segments on the neighboring interfaces.
C. Release of the dislocation from the particle corner and the annihilation of the released
segments.
Loading: 150 MPa, along y =[010]
10
Modeling conditions
Dislocation properties
• 12 straight ½ a0<110>{111} mixed dislocations in
the γ channels ( = 6.7×1013 m-2).
• Edge and screw dislocation mobility: 1 Pa-1s-1
• Glide/climb mobility ratio: Mg/Mc = 10
Simulation box
• 0.42 μm cuboidal γ´ precipitates are aligned in
<100> directions.
• The γ channel width is 80 nm and γ´ phase fraction
is 59%.
• Loading conditions: 150 to 350 MPa applied along
[100]
[100] loading
11
Effect of loading conditions on creep rate
M. Kolbe, A. Dlouhy,
G. Eggeler, MSEA, 1998.
S.M. Hafez Haghighat,
G. Eggeler, D. Raabe,
Acta Materialia (2013)
T.M. Pollock, A.S. Argon, Acta
Metall Mater, 40 (1992) 1-30
12
What initial dislocation microstructure
should be considered in DDD simulation?
CMSX-4
13
Experimental procedure
ABD-2: Ni -8Cr -10Co -3Re -8.5W -5.9Al -8.5Ta (wt.%)*
Creep test: 900°C , 450 MPa along [001] deformed up to fracture
Analyzing cross section
Surface orientation:
(001)
Approachable reflectors:
(200) (0-20) (220) (2-20)
Picture in courtesy of M. Nellesen *R.C. Reed et al. / Acta Materialia 57 (2009) 5898
Sample
1 2 3 4 5 6
Analysed surfaces
14
Electron Channeling Contrast Imaging (ECCI)
x y
z
channeling
x z
y
back-
scatterin
g
• ECCI imaging due to stacking faults
• Visibility of lattice defects at 30 kV
• Defects are seen up to 80-100 nm below
the surface
Stacking faults observed in TWIP steels and trace analysis.
S. Zaefferer, N.N. Elhami, Acta Mat. 75 (2014) 20
15
b1 : screw
b2 , b3 :60° mixed
b4 :pure edge
H. Gabrisch et al. 1996 Z. P. Luo et al. 2003 T. M. Pollock 2002
Characterizing dislocation line direction
on {100} interfaces
Tan(α) = Dv / L
At 30 kv Dv is approximately 90 nm
Beam
Normal
Sample
surface
Visual length (L)
Inclination angle (α)
Depth of visibility
(Dv)
[100]
[010]
[001]
σ
b1
b2
b3
b4
Dislocation Characterization using ECCI
16
Dislocation Characterization using ECCI
Surface orientation near (001) pole
• Dislocations become (almost)
invisible when g∙b = 0 (g: diffraction
vector, b: Burgers vector)
g=(200) g=(2-20)
g=(0-20) g=(220)
b1
b1
b3
Max Planck Institute for Iron Research, Düsseldorf,
Germany zhangqi wang 18
Outcome of ECCI investigations
• In the undeformed material dense arrays/networks of dislocations
are observed at the dendrite interfaces being of {110} or {010} type.
• At early stages of creep dislocation dipoles are emitted from the
dendrite interfaces into the γ channels.
• ECCI analyses indicate that the majority of glissile dislocations are
of b= ½ a0<110>.
19
undeformed sample 0.3 % creep strain undeformed sample
(010)
(101)
New modeling conditions
• 27 γ´ particles (0.42 μm) oriented along <100> directions
• γ channel width: 80 nm
• γ´ phase fraction: 59%
• Loading conditions: 450 MPa along dendrite column
• Boundaries contain 1 to 3 mixed dislocations per channel
with b = ½ a0 [011]
• Two different boundaries are investigated:
(010) and (1-10)
20
Dislocation propagations from interdendrite boundary
21
450 MPa loading along the boundary, 3 dislocations/channel
Simulated creep rate
• Two stages for creep process are seen:
– Initial low rate stage; dislocation generation & plastic strain are consistent
– Following high rate stage; dislocation generation & creep rate
22
Dislocation multiplication rate
23
Misorientation angle (d at boundary)
• At high misorientation angle (high d at interdendrite boundary) creep rate is
higher.
• However, the rate of dislocation multiplication is higher at low misorientation
angle
• Contribution of the (010) and (101) boundaries to creep plastic strain is
comparable
24
Low vs high temperature regime
• At low temperature plastic strain is small due to filling of channels by
dislocations, where dislocation back forces prevent further dislocation
movements.
• In the absence of climb process the dislocation multiplication rate is notability
higher than that of high temperature creep, where the climb mobility is active.
25
26
Summary
For single crystal Ni base superalloys:
1. Discrete dislocation dynamics (DDD) is used for studying the mobility and
multiplication process of dislocations to characterize mechanisms of
dislocation creep process.
2. Electron Channeling Contrast Imaging (ECCI) technique is used to
characterize the GNDs composing the interdendrite boundaries.
3. The role of interdendrite boundaries as the source of the necessary dislocation
content has been clarified using the DDD simulations.
4. The effect of dislocation density (misorientation angle) at the interdendrite
boundary on creep rate is minor.
5. Dependence of creep rate to the character of interdendrite boundary, (010) vs.
(1-10), is not remarkable.
6. Dislocation features seen in the simulated microstructures would interpret the
experimentally observed dislocation configurations.
27
Thanks for your attention